[CASSANDRA-11571] Optimize the overlapping lookup, by calculating all the bounds in advance.

• Type: Improvement
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 3.6
• Component/s: Local/Compaction
• Assignee: Dikang Gu
• Reporter: Dikang Gu

Authors

Dikang Gu

Reviewers

Marcus Eriksson

• Created: 13/Apr/16 22:40
• Updated: 16/Apr/19 09:30
• Resolved: 19/Apr/16 11:39

Description

When L0 sstable backs up (because of repair or other reasons), I find that a lot of CPU is using to construct the Bounds.

"CompactionExecutor:223" #1557 daemon prio=1 os_prio=4 tid=0x00007f88f401d800 nid=0x2303ab runnable [0x00007f824d735000]
   java.lang.Thread.State: RUNNABLE
        at org.apache.cassandra.dht.AbstractBounds.strictlyWrapsAround(AbstractBounds.java:86)
        at org.apache.cassandra.dht.Bounds.<init>(Bounds.java:44)
        at org.apache.cassandra.db.compaction.LeveledManifest.overlapping(LeveledManifest.java:533)
        at org.apache.cassandra.db.compaction.LeveledManifest.overlapping(LeveledManifest.java:520)
        at org.apache.cassandra.db.compaction.LeveledManifest.getCandidatesFor(LeveledManifest.java:595)
        at org.apache.cassandra.db.compaction.LeveledManifest.getCompactionCandidates(LeveledManifest.java:349)
        - locked <0x00007f8e11e67900> (a org.apache.cassandra.db.compaction.LeveledManifest)
        at org.apache.cassandra.db.compaction.LeveledCompactionStrategy.getNextBackgroundTask(LeveledCompactionStrategy.java:97)
        - locked <0x00007f8e11b1d780> (a org.apache.cassandra.db.compaction.LeveledCompactionStrategy)
        at org.apache.cassandra.db.compaction.WrappingCompactionStrategy.getNextBackgroundTask(WrappingCompactionStrategy.java:78)
        - locked <0x00007f8e110931a0> (a org.apache.cassandra.db.compaction.WrappingCompactionStrategy)
        at org.apache.cassandra.db.compaction.CompactionManager$BackgroundCompactionCandidate.run(CompactionManager.java:250)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
        at java.util.concurrent.FutureTask.run(FutureTask.java:266)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
        at java.lang.Thread.run(Thread.java:745)

From the code, we may construct the bounds multiply times, my patch optimizes it by calculating it in advance.